Expression of p16 induces transcriptional downregulation of the RB ...

Oncogene (1998) 16, 1 ± 8  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00

Expression of p16 induces transcriptional downregulation of the RB gene Xianjun Fang, Xiaomei Jin1, Hong-Ji Xu, Lin Liu2, Hong-Qi Peng, David Hogg2, Jack A Roth1, Yinhua Yu, Fengji Xu, Robert C Bast, Jr and Gordon B Mills Department of Molecular Oncology, Division of Medicine; 1Department of Thoracic and Cardiovascular Surgery, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030, USA; 2Department of Medicine, University of Toronto and Toronto General Hospital, M5G, 1L7, Canada

The RB and p16INK4A tumor suppressor genes function in the same pathway of cell cycle control. Previous evidence indicates that the p16INK4A gene is transcriptionally repressed by the RB gene product, pRB. In this study using human ovarian cancer cell lines, we found that RB protein and mRNA were expressed at higher levels in cell lines lacking p16 than in those with normal p16. Since this suggests a potential role of p16 in regulating the cellular level of pRB, we studied the eect of wild-type p16INK4A on expression of the RB gene. Introduction of p16INK4A, carried by an adenovirus vector, into p16negative cell lines dramatically decreased expression of RB protein and mRNA. Nuclei run-o assays demonstrated that p16 expression induced transcriptional downregulation of the RB gene. These results indicate that expression of RB is inversely regulated by p16. The ®ndings reveal a new dimension of pRB-p16 interaction and should have implications for p16INK4A-mediated gene therapy. Keywords: p16; RB; expression; ovarian cancer

Introduction Many human tumors contain abnormalities in one or more of the genes responsible for regulating cell cycle progression. These include the well-characterized tumor suppressor genes, RB and p53. The product of the RB gene, pRB, functions to prevent the cell from entering S phase, whereas phosphorylated or mutated forms of the protein are incapable of arresting the cell in G1 (Goodrich et al., 1991; Xu et al., 1991a). Phosphorylation of pRB is mediated by complexes comprised of a D-type cyclin and cyclin-dependent protein kinases (CDK4)/(CDK6) (Weinberg, 1995; Kamb, 1995). The activity of these kinases is in turn negatively regulated by cyclin kinase inhibitors including p16 encoded by p16INK4A (MTS1) (Serrano et al., 1993). Thus the cascade composed of pRB, cyclin D1, CDK4/CDK6 and p16 plays a central role in cell cycle control. Deregulation of individual components of this pathway has been implicated in various human malignancies (Weinberg, 1995). For instances, in small cell lung carcinomas, sarcomas and bladder carcinomas, pRB function is frequently lost through mutations of the RB

Correspondence: GB Mills Received 19 September 1996; revised 22 August 1997; accepted 26 August 1997

gene (Horowitz et al., 1990). In many oesophageal, breast, and squamous cell carcinomas, the cyclin D1 gene is ampli®ed (Jiang et al., 1992; Lammie et al., 1991). Cyclin D1 overexpression is achieved in B cell lymphomas through chromosomal translocation (Motokura et al., 1991). The cdk4 gene is ampli®ed in many glioblastomas and some gliomas or cell lines (He et al., 1994, 1995; Schmidt et al., 1994). p16INK4A has been found deleted or mutated in high percentage of cell lines of melanomas, lymphomas, and breast, ovary, bladder, and kidney carcinomas (Kamb et al., 1994; Nobori et al., 1994). Although with lower incidence, the p16 gene is also found to be deleted or mutated in uncultured oesophageal squamous cell carcinomas (Mori et al., 1994; Zhou et al., 1994), glioblastomas (Schmidt, et al., 1994; Jen et al., 1994; Ueki, et al., 1996), lung (Okamoto et al., 1994), bladder (Spruck et al., 1994), and pancreatic carcinomas (Caldas et al., 1994). The centrality of this control pathway and the importance of its deregulation during tumor progression are further highlighted by studies showing that, in small cell lung cancer (Otterson, et al., 1994; Shapiro et al., 1995a; Kelley et al., 1995), and glioblastomas (Ueki et al., 1996), the expression of functional p16 and the pRB is inversely correlated. p16INK4A is deleted or mutated only in tumors or cell lines with normal pRB. In contrast, where the function of pRB is lost through mutations or is blocked by viral oncoproteins such as the SV40 T antigen and HPV E7, p16INK4A is usually unaected and expressed at elevated levels. This suggests that losses of p16 and pRB are functionally analogous in oncogenesis. The correlation of RB mutation with enhanced expression of p16 also indicates a role for normal pRB in the control of p16 expression. Indeed, Li et al. (1994) have demonstrated that, in cultured cells, pRB represses transcriptional activity of the p16INK4A gene promoter. In many other types of human malignancies, such as breast and ovarian cancers, the p16INK4A and RB genes are rarely aected (Weinberg, 1995). Nevertheless, the p16INK4A gene, but not the RB gene, is deleted or mutated in high percentages of tumor-derived cell lines (Xu et al., 1994; Rodabaugh, et al., 1995). In this study, we have observed that ovarian cancer cell lines lacking p16 express RB protein and mRNA at higher levels than those with normal p16. This inverse correlation between RB expression and loss of p16 strongly suggests that p16 expression may restrain that of the RB gene. We herein present evidence that p16 expression is associated with decreased transcription of the RB gene.

p16 downregulates RB transcription X Fang et al

A2780

DOV 13

OVCA 420

OVCAR-3

HOC-1

OCC.1

OVCA 432

OVCA 429

OVCA 433

CaOV-3

SKOV-3

Saos-2

HOC-1

CaOV-3

— pRB

A2780

DOV 13

OVCA 420

OVCAR-3

HOC-1

OCC.1

OVCA 432

Figure 1 Expression of pRB and p16 proteins in ovarian cancer cell lines. Cells were grown to approximately 80% con¯uence. Lysates were prepared for Western blot analyses. Equal amounts of cellular protein from each cell line were subjected to SDSpolyacrilamide gel electrophorosis (8% for pRB and 12% for p16). Multiple bands representing dierent states of phosphorylation of pRB and a single one for p16 are indicated at the right of the panels. The bottom panel shows the presence of pRB in CaOV-3-3 and HOC.1 (longer exposure) with Saos-2 as a negative control

OVCA 429

Expression of pRB and p16 proteins in the ovarian cancer cell lines was analysed by Western blotting. Consistent with the result of immunohistochemical staining, pRB of approximately 110 ± 114 Kd was present in all lines (Figure 1, top). There was considerable variation in the expression levels of pRB among these cell lines. The lowest expressor was CaOV-3 which exhibited visible bands only after prolonged exposure (Figure 1, bottom). These cell lines showed two or multiple bands of pRB protein, apparently representing dierent extents of phosphorylation of pRB (Chen et al., 1989). In HOC-1, the hypophosphorylated pRB was predominant over the

— p16

OVCA 433

Expression of pRB and p16 proteins

— pRB

CaOV-3

To study the relationship between p16 and pRB expression, we collected 12 cell lines derived from ovarian carcinomas. The RB status of these lines was determined by histochemical staining with an antibody (RB-WL-1) which reacts with both hypophosphorylated and hyperphosphorylated forms of the RB protein (Xu et al., 1991b). Mutations of the RB gene usually result in lack of RB protein or abnormal protein localization in the cell (Xu et al., 1991a,b). Consistent with previous ®ndings in ovarian cancer tissues (Dodson et al., 1994), all the ovarian cancer cell lines showed strong RB staining in nuclei with the exception of CaOV-3 which contained cells with faintly stained nuclei (Table 1). As homozygous deletion is the major form of mutation for p16INK4A (Kamb et al., 1994; Nobori et al., 1994), the status of p16INK4A in the lines was examined by PCR ampli®cation of genomic DNA using speci®c primers for the p16INK4A exon 1 and 2. As summarized in Table 1, six of the 12 cell lines (HEY, SKOV-3, OVCA 433, OVCA 429, OCC.1, DOV 13) showed homozygous deletion of p16INK4A as re¯ected by the absence of PCR products of the exon 1, 2 or both sequences. The status of RB and p16INK4A in these ovarian cancer cell lines was con®rmed by Western and Northern blot analyses (see Figures 1 and 2).

HEY

The status of the RB and p16INK4A genes in ovarian cancer cell lines

hyperphosphorylated forms of the protein. Since the cell line had a prolonged doubling time (Buick et al., 1985), this could be due to a high percentage of G0/G1 cells present in the unsynchronized culture although the possibility can not be excluded that this line and/or other lines we examined here may carry a mutation in the RB gene. p16 expression was examined in the same lysates (Figure 1, middle). The six cell lines with deleted p16INK4A did not show p16 protein. Among the other six cell lines, only four (CaOV-3, OVCA 432, HOC-1 and

SKOV-3

Results

HEY

2

— RB 9.2 9.5 0.3 7.0 5.7 1.9 7.1 4.3 3.2 7.1 5.9 5.9

Table 1 The status of the RB and p16INK4A gene in ovarian cancer cell lines Cell lines HEY SKOV-3 CaOV-3 OVCA 433 OVCA 429 OVCA 432 OCC.1 HOC-1 OVCAR-3 OVCA 420 DOV 13 A2780 a

RBa

p16b

+ + + + + + + + + + + +

± ± + ± ± + ± + + + ± +

Histochemical staining for RB protein in nuclei; +, positive; +, weakly positive; bPCR ampli®cation of p16INK4A exon I and II; +, detection of both exons; ±, lack of one or both exons

— p16INK4A

— 18S Figure 2 Expression of RB and p16INK4A mRNA in ovarian cancer cell lines. Total cellular RNA was extracted from cells cultured under the same conditions as in Figure 1. Northern blotting was performed using 32P-labeled DNA sequences of the human RB cDNA, p16INK4A exon 1, and 18S rRNA as probes. The locations of RB and p16iINK4A transcripts and 18S RNA are indicated at the right of the panels. The numbers under each lane of the RB panel (top) represent relative intensities of RB mRNA bands after normalization to those of 18S bands, as determined by densitometry


3

Ad-p16

SKOV-3 Ad-Lac Z

Ad-p16

Ad-Lac Z

OVCA 429 Ad-p16

HEY Ad-Lac Z

OVCAR-3) expressed detectable quantities of the p16 protein. Another two cell lines (OVCA 420 and A2780) did not show detectable levels of p16 protein despite normal PCR ampli®cation of p16INK4A exon 1 and 2 (see Table 1). When compared with the pRB expression pattern, three of the four p16 protein-positive cell lines (CaOV-3, OVCA 432, HOC-1) contained the lowest amounts of pRB among all the cell lines, suggesting that p16 may play a role in regulating pRB expression. However, the fourth p16-positive cell line, OVCAR-3, seems to be an exception in that it did not show marked reduced levels of pRB compared to p16negative cell lines. Thus p16 may not be the only factor in regulating the cellular level of pRB.

— pRB

— p16

Expression of RB and p16INK4A mRNA We next performed Northern blot analysis to determine the steady-state levels of RB and p16INK4A mRNA in the ovarian cancer cell lines. As demonstrated in Figure 2, all twelve cell lines contained readily detectable levels of the 4.8 Kb RB transcript except CaOV-3 which had an extremely low level that required extended exposure for detection (not shown). The relative intensity of the RB mRNA bands were largely consistent with that of RB protein signals shown in Figure 1. The four p16 protein-positive cell lines (CaOV-3, OVCA 432, HOC-1 and OVCAR-3) all displayed p16INK4A transcripts (Figure 2, middle). Additionally, marginal expression of p16INK4A mRNA was seen in A2780 and no signal was detected in OVCA 420. Both of these cell lines lacked detectable p16 protein, suggesting that, in addition to homozygous deletion, p16INK4A expression could be blocked through point mutations and transcriptional or translational/post-translational regulation in ovarian cancer cell lines. Multiple pathways leading to p16 inactivation have been described in other types of tumor cells (Shapiro et al., 1995b; Gonzalez-Zulueta et al., 1995). Once again, the four p16 protein-positive cell lines (CaOV-3, OVCA 432, HOC-1, and OVCAR-3) were among those showing the lowest abundance of RB transcript (Figure 2). These results suggests a negative regulatory eect of p16 protein on the transcriptional activity of the RB gene promoter. Eect of introduced p16INK4A on RB expression To explore the possibility that p16 may restrain expression of RB, we introduced wild-type p16INK4A into p16-negative ovarian cancer cell lines by infection with recombinant adenovirus harboring the human p16INK4A cDNA, Ad-p16. This virus-mediated system yielded a high infectivity in human cancer cell lines as previously demonstrated (Jin et al., 1995). As recipient cells, HEY, OVCA 429 and SKOV-3 were chosen to represent cell lines with a homozygous deletion of p16INK4A. In all three cell lines, Western blot analysis showed that p16 was highly expressed in Ad-p16infected but not Ad-Lac Z-infected cells (Figure 3). Infection with Ad-p16 resulted in a marked decrease in the level of total RB protein compared with infection with Ad-Lac Z. This change was primarily derived from the disappearance of hyperphosphorylated pRB with no dramatic change in levels of hypophosphorylated form.

— β-actin

Figure 3 The eect of introduced p16INK4A on expression of RB protein in ovarian cancer cell lines. Three p16-negative cell lines, HEY, OVCA 429 and SKOV-3, were infected with Ad-Lac Z and Ad-p16. Expression of pRB, p16 and b-actin proteins in these cells was examined by Western blotting analyses as described in Figure 1

Northern blot analysis of total cellular RNA was then performed to determine if a corresponding alteration in RB mRNA levels could account for the observed decrease in RB protein. As expected, cells infected with Ad-p16 displayed a decrease in RB mRNA levels (Figure 4a). Thus p16 expression is indeed associated with a downregulation of RB mRNA expression. It seems unlikely that this downregulation is due to pRB-mediated autorepression since, compared to Ad-Lac Z-infected cells, those infected with Ad-p16 did not possess more hypo-phosphorylated pRB (Figure 3) which is required for autorepression (Shan et al., 1994; Gill et al., 1994; Bremner et al., 1995). To examine the generality of this regulation, we performed a similar experiment in a bladder carcinoma cell line, T24, which is known to lack p16 protein (Parry et al., 1995). T24 cells infected with Ad-p16 also showed dramatically decreased RB mRNA levels (Figure 4b), suggesting that the eect of p16 on RB expression was not restricted to epithelial ovarian cancer cell lines. Cell cycle-independent eect of p16 Expression of p16 has been shown to induce cell cycle arrest at G1. The eect of p16 on expression of RB could re¯ect the accumulation of cells in the G1 phase of the cell cycle, as, in some cell types, G1- phase cells contain less RB protein than do cells in S and G2 phases (Xu et al., 1991a; Shan et al., 1994). Although not directly demonstrated, it has been assumed that this cell-cycle regulation is controlled at the transcriptional level. To address a possible cell cycle-related eect of p16, we examined RB mRNA expression in OVCA 420 cells which contain nearly 80% cells at G0/


+TGF-β

Ad-p16

T24 Ad-Lac Z

Ad-p16

Ad-Lac Z

Ad-p16

Ad-Lac Z

Ad-p16

Ad-Lac Z

OVCA 429 SKOV-3

–TGF-β

b HEY

Ad-p16

a

Ad-Lac Z

4

— RB — RB

— β-actin

Transcriptional repression of RB in p16-expressing cells To establish whether the observed decrease in RB mRNA was due to a decrease in the transcription rate of the RB gene, we conducted a nuclear run-o assay in SKOV-3 and HEY cells. As seen in Figure 6, nuclei isolated from Ad-p16-infected cells synthesized a signi®cantly lower amount of RB transcripts than nuclei from control cells, as re¯ected by dierent intensities of hybridizing bands. As a control, the rate of transcription of the b-actin gene was not signi®cantly altered in Ad-p16-infected cells. We also performed experiments to examine the stability of RB mRNA with no detectable change seen in Ad-p16-

78.2 9.4 12.0

90.9 3.0 6.3

Figure 5 Cell cycle-independent eect of p16INK4A on RB mRNA expression in OVCA 420 cells. Total cellular RNA was isolated from cells infected with Ad-Lac Z or Ad-p16 (left) and from cells treated with vehicle (7TGF-b) or TGF-b (+TGF-b) (5 ng/ml) (right). RB mRNA expression levels were examined by Northern blot analyses using the same conditions as in Figure 2. Cell cycle distributions in each treatment were determined by ¯ow cytometry. Percentages of G0/G1, S and G2/M phases are shown below the corresponding lanes. Consistent results were obtained from two independent experiments

SKOV-3

HEY Ad-p16

G1 phase under normal culture conditions. This cell line did not express detectable p16 protein (see Figure 1). Following infection with Ad-p16, 90% of cells were in G0/G1 phase, indicating an increment of 10% in G1 (Figure 5). Despite inducing only a slight increase in G0/G1 population, Ad-p16 infection was accompanied with a signi®cant reduction in the level of RB mRNA. We compared the eect of p16 with that of TGF-b in OVCA 420 cells which are highly sensitive to the growth inhibitory action of TGF-b (Berchuck et al., 1992). Both p16 and TGF-b inhibit cell cycle progression through preventing phosphorylation of pRB (Lukas et al., 1995; Laiho et al., 1990). After incubation of OVCA 420 cells with 5 ng/ml TGF-b for 42 ± 48 h, G0/G1 cells were increased from approximately 80% to 90%, comparable to the shift induced by infection with Ad-p16. These TGF-b-treated cells, however, did not show a signi®cant change in RB mRNA levels (Figure 5). Although the lack of a response to TGF-b has precluded a similar experiment in other p16-negative cell lines such as SKOV-3 and HEY (Berchuck et al., 1992; ), the result derived from OVCA 420 indicates that, in this cell line, p16 induces downregulation of RB mRNA largely through a cell cycle-independent mechanism.

G0/G1 S G2/M

Ad-Lac Z

Figure 4 The eect of introduced p16INK4A on expression of RB mRNA in ovarian (HEY, OVCA 429, SKOV-3) and bladder (T24) carcinoma cell lines. Total cellular RNA was extracted from Ad-Lac Z and Ad-p16-infected cells and subjected to Northern blot analyses using 32P labeled cDNAs of the human RB, p16INK4A, and b-actin genes as probes

91.1 3.4 5.3

Ad-p16

— β-actin

79.7 10.7 9.6

Ad-Lac Z

— p16INK4A

— β-actin

— β-actin — RB

— RB Figure 6 Transcriptional downregulation of the RB gene in p16expressing cells. Nuclear run-o transcription was conducted in nuclei isolated from SKOV-3 and HEY cells infected with Ad-Lac Z or Ad-p16. Lysates were made from parallel cultures to examine the expression of pRB and p16 with results demonstrated in Figure 3. Autoradiogram shows amounts of radiolabeled RNA synthesized by the nuclei that hybridized to plasmids containing cDNAs of the human b-actin and RB genes. Three independent experiments were performed with similar results

infected cells compared with those infected with AdLac Z (data not shown). Thus the decreased RB mRNA levels associated with p16 expression likely resulted primarily from decreased transcription of the RB gene. The eect of p16INK4A on E2F-1 expression The expression of E2F-1 is stimulated by cyclin D1, an action which can be inhibited by p16 (Johnson, 1995). E2F-1 has been shown to positively regulate expression of the RB gene (Shan et al., 1994; Gill et al., 1994). To ask whether E2F-1 could play a role in the negative regulation of expression of RB, we examined the eect of p16 on the levels of E2F-1. Infection with Ad-p16 induced a dramatic decrease in the level of E2F-1 protein in all three p16-negative cell lines (Figure 7a).


5

b

Ad-p16

SKOV-3 Ad-Lac Z

Ad-p16

Ad-Lac Z

Ad-p16

OVCA 429 Ad-p16

HEY

SKOV-3 Ad-Lac Z

Ad-p16

Ad-Lac Z

OVCA 429 Ad-p16

Ad-Lac Z

HEY

Ad-Lac Z

a

— E2F-1

— E2F-1

— β-actin

— β-actin

Figure 7 The eect of p16INK4A on E2F-1 Expression. Lysates and total cellular RNA were prepared from SKOV-3, OVCA 429 and HEY cells infected with Ad-Lac Z or Ad-p16. Expression of E2F-1 protein (a) and mRNA (b) were examined with b-actin as controls for comparison. The procedures for Western and Northern blot analyses are the same as described in Figures 1 and 2. The human E2F-1 cDNA was used as probe for Northern blot analysis

A concomitant decrease in the level of E2F-1 mRNA was observed (Figure 7b). Thus p16 expression is associated with downregulation of E2F-1 activity. Discussion Homozygous deletion of p16INK4A has been detected in high percentages of cell lines derived from all major forms of human malignancies. In the ovarian cancer cell lines we have studied here, p16INK4A deletion occurred at a rate of 50% in spite of the lack of such a frequency of mutation in primary ovarian cancer cells (Rodabaugh et al., 1995). Since they ubiquitously contained intact RB, these cell lines oer an ideal cellular model to assess the in¯uence of p16 on the expression of RB. By showing that RB expression is transcriptionally repressed in the presence of p16, we have herein identi®ed a new dimension of crosstalk between pRB and p16. Combined with the observation that pRB suppresses p16INK4A expression (Li et al., 1994), our result suggests that expression of RB and p16INK4A is mutually counterbalanced. Through such a bi-directional feedback, expression of these two growth-inhibitory genes, which act on the same regulatory pathway, would be strictly and precisely controlled. A change in the expression of one gene would be compensated for by an opposite alteration in the other gene so that the total growth-inhibitory activity of this pathway is maintained. If the function of one partner is abrogated through mutation or deletion, the other would be overexpressed as seen in some tumors or tumor cell lines. Under these circumstances, the high expression of a single gene (either RB or p16INK4A) would be incapable of inhibiting cell cycle progression. By binding to CDK4 and CDK6, p16 prevents these two kinases from phosphorylating pRB, likely resulting in an transient accumulation of the hypophosphorylated state of the RB protein. This change would probably simultaneously trigger downregulation of RB transcription through autorepression (Shan et al., 1994; Gill et al., 1994; Bremner et al., 1995). However, once hypophosphorylated pRB is equilibrated through transcriptional control, to the level not higher than that in control cells, pRB-mediated

autorepression would no longer be functional. In our study, Ad-p16-infected cells did not contain more hypophosphorylated pRB than did Ad-Lac Z-infected control cells (Figure 3). Thus the lower transcription rate of the RB gene in these p16-expressing cells (Figure 4) is not accomplished solely by autorepression. Other alternative or cooperative mechanisms may be involved. Hints as to a possible pathway by which p16 could regulate RB transcription may be suggested by recent studies on the eects of cyclin D1 and p16 on E2F-1mediated transcription (Oswald et al., 1994; Schulze et al., 1994; Hengstschlager et al., 1996). In these experiments, cyclin D1 expression signi®cantly induced transcription from promoters containing E2F-1 binding sites such as those of the dihydrofolate reductase gene, c-myc and the thymidine kinase gene. It is postulated that cyclin D1-dependent kinase activity stimulates phosphorylation of pRB and other related proteins (p107 and p130), a process leading to release of E2F-1 and activation of E2F-1-mediated transcription. Johnson (1995), however, described an alternative mechanism, that is, stimulation of E2F-1 gene expression by cyclin D1. As a strong and speci®c inhibitor of D-type cyclin kinase activity, p16 expression has been demonstrated to block these actions of cyclin D1 (Schulze et al., 1994; Johnson, 1995; Hengstschlager et al., 1996). The RB gene promoter contains a potential binding site for E2F-1 (Shan et al., 1994; Gill et al., 1994). To explore the possibility that deregulated E2F-1 activity may play a role in p16-mediated repression of the RB gene, we examined the expression of E2F-1 in Ad-p16-infected cells and found a marked decrease in E2F-1 mRNA and protein. If E2F-1 is a rate-limiting factor in controlling the activity of the RB gene promoter, reduced E2F-1 activity in p16-expressing cells could conceivably limit RB transcription. To assess the contribution of p16-induced G1 accumulation to the observed downregulation of RB transcription, we compared the eects of p16 infection and TGF-b treatment in OVCA 420 cells. Both agents induced similar degree of cell cycle arrest with 90% of cells in G1. The fact that potent inhibition of expression of RB mRNA was only seen in Ad-p16infected cells indicates a cell cycle independent pathway


6

mediating the downregulation of RB expression by p16. However, it remains possible that, in other cell lines, the eect of p16 on cell cycle transit may contribute to the observed downregulation of the RB gene. In a search for changes in the expression of other genes in Ad-p16-infected cells, we found that, in contrast to pRB, CDK4 and a protein related to p202 (Lengyel et al., 1995) were expressed at elevated levels (unpublished data). The elevated CDK4 level likely re¯ects an enhanced stability of the protein on binding with p16. The reason for and the signi®cance of the change in the p202-related protein is not clear. As proteins of the p202 family have been implicated in inhibition of E2F-1 activity (Lengyel et al., 1995) which seems to be important in controlling transcription of the RB gene (Shan et al., 1994), the change in the p202-related protein could also contribute to the observed downregulation of RB expression in p16expressing cells. Whatever the mechanism, p16-mediated suppression of RB gene transcription provides a novel and potentially useful model for studying the molecular mechanisms executing the negative regulation of RB expression. The ®ndings of the present study should also have implications for p16INK4A-mediated gene therapy (Jin et al., 1995). The tumor-suppressing activity of p16 is achieved through its eector, pRB. The tendency of recipient tumor cells to reduce RB transcription and protein levels in response to introduced p16 may functionally compromise the long-term eect of p16. To obtain an optimal tumorsuppressing activity of p16INK4A-mediated gene therapy, it may be desirable to co-introduce a wild-type RB under the control of a foreign promoter which is not aected by p16.

Materials and methods Cells Twelve ovarian cancer cell lines (HEY, SKOV-3, CaOV-3, OVCA 433, OVCA 429, OVCA 432, OVCA 420, DOV 13, OVCAR-3, OCC. 1, HOC-1, and A2780), one osteosarcoma line (Saos-2) and one bladder carcinoma line (T24) were collected and utilized in this study. SKOV-3, CaOV-3, OVCAR-3, Saos-2 and T24 were obtained from ATCC and propagated in culture as recommended by the supplier. HEY and HOC-1 lines were kindly provided by Dr R Buick (University of Toronto, Toronto, Ont, Canada) and cultured as previously described (Buick et al., 1985). The line A2780 was a gift of Dr TC Hamilton (Fox Chase Cancer Center, Philadelphia, Penn). The sources and maintenance of other ovarian cancer cell lines have been previously described (Bast et al., 1983; Pales et al., 1993). Immunohistochemistry Cells of ovarian cancer cell lines were grown on sterile coverslips in tissue culture dishes until 60 ± 80% con¯uent. The coverslips were washed, ®xed and stained for pRB using poly-clonal anti-pRB antibody, RB-WL-1, as described by Xu et al. (1991b). A cell line was considered to be RB+ if any of its cells had RB nuclear staining whereas a line was de®ned as RB7 only if all cells lacked RB nuclear staining. Cell lines T24 and HTB9 were included in each staining experiment as positive and negative control, respectively.

PCR Genomic DNA was extracted from cultured cells using standard methods (Peng et al., 1993). Two pairs of ¯anking primers were used to amplify exon 1 and 2 of the p16INK4A gene (primers for exon 1: 5'-TTC GGA GAG GGG GAG AAC AG/3'-AAG TTC GTC CTC CAG AGT CG, and primers for exon 2: 5'-TGG CTC TGA CCA TTC TGT TC/3'-TTT GGA AGC TCT CAG GGT AC). PCR reactions were carried out in a buer containing 30 ± 50 ng of genomic DNA, 2.0 mM MgCl2, 0.2 mM each of dNTP, 10 mM Tris HCl (pH 8.6), 50 mM KCl, 5% DMSO, 0.4 mM of each primer and 2.5 units of Taq polymerase for 35 cycles. A negative control (water) was run in parallel in each experiment. Western blotting Subcon¯uent cells in 100 mm dishes were lysed with 1.2 ml of ice-cold lysis buer [50 mM Tris-HCl (pH 8.0), 100 mM NaCl, 0.2% NP-40, 0.2% sodium deoxycholate, and 0.1% SDS] containing 50 mg ml71 aprotinin, 1 mM Na vanadate, and 25 mM NaF]. Lysates were transferred to microcentrifuge tubes, vortexed and passed through a 23 g needle several times. Following a 10 min incubation on ice, the lysates were clari®ed by centrifugation for 5 min at full speed. Protein concentrations were determined using BCA kit (PIERCE, Rockford, Ill). Equal amounts of total cellular protein were separated by SDS-polyacrylamide gel electrophoresis, transferred to immobilon (PVDF), and immunoblotted with a mouse monoclonal antibody against the human pRB (UBI), with a rabbit anti-p16 antibody (Santa Cruz), or with a rabbit anti-E2F-1 antibody (Santa Cruz) according to protocols provided by the manufacturers. Immunocomplexes were visualized by an enhanced chemiluminescence detection kit (Amersham) using horseradish peroxidase-conjugated secondary antibodies. Northern blotting Total cellular RNA was extracted from cultured cell lines using the guanidinium isothiocyanate-phenol chloroform method (Chirgwin et al., 1979). RNA samples (15 mg) were size-fractionated by formaldehyde/agarose-gel electrophoresis, stained with ethidium bromide, and transferred to N+ hybrid nylon. RNA was immobilized by u.v. crosslinking, and then prehybridized and hybridized to 32P-labelled cDNA probes in 50% formamide, 66SSC, 106Denhardt's solution, 10 mM EDTA, 0.1% SDS and 150 mg/ml denatured salmon sperm DNA. cDNA clones for the RB, p16INK4A, E2F-1 genes were gifts of Dr W Benedict (Balor College of Medicine, Houston, TX), Dr D Beach (Cold Spring Harbor Laboratories), and Dr D Choubey (University of Texas MD Anderson Cancer Center), respectively. Quality of RNA samples was con®rmed by re-hybridization of nylon membranes to the human b-actin cDNA or to the DNA sequence for 18S rRNA obtained from the ATCC. Recombinant adenovirus and infection of cell lines The details for generation of recombinant p16INK4A adenovirus, Ad-p16 have been previously described by Jin et al (1995). Another recombinant adenovirus, Ad-Lac Z, which carries the Lac Z gene of Escherichia coli in place of p16INK4A was used as a control in the experiments. Individual clones of the Ad-p16 and Ad-Lac Z viruses were obtained by plaque puri®cation and propagated according to Graham and Prevec (1991). The titer of virus stock was determined by plaque assays. Cell lines were grown to approximately 60% con¯uence before infection with viruses at a titer of approximately 10 p.f.u./cell. The viruses, diluted in an appropriate volume of culture medium


(2.5 mls for a 100 mm plate), were overlaid on cell monolayers and incubated for 1 h at 378C with brief agitation every 15 min. Culture medium was then brought up to 10 mls for a 42 ± 48 h incubation. Flow cytometry Cells infected with Ad-p16 or Ad-Lac Z, or treated with TGF-b (GIBCO/BRL) or vehicle were harvested by trypsinization, washed in PBS, and ®xed in 1% paraformadehyde. Cells were stained with propidium iodide before ¯ow cytometric analysis of cell cycle status with a FACScan (Becton Dickinson). Nuclear run-o transcription Cells in 100 mm dishes were washed twice with cold PBS, scraped with a rubber policeman, and centrifuged at 500 g for 5 min. The cell pellets were resuspended in lysis buer [10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl2, 0.5% NP-40], incubated for 5 ± 10 min on ice, and centrifuged again. The pellets of nuclei was washed once with the lysis buer, resuspended in 50 mM Tris-HCl (pH 8.3), 40% glycerol, 5 mM MgCl2 and 0.1 mM EDTA,

and stored at 7808C. For labelling, thawed nuclei (16107) were resuspended in transcription buer containing 0.5 m M each of ATP, GTP and CTP, 5 mM Tris-HCl (pH 8.0), 2.5 mM MgCl2, 1 mM MnCl2 and 150 mM KCl in the presence of 250 mCi [32P]UTP at 308C for 30 min. Reaction was stopped by addition of solution D [6 M guanidinium isothiocyanate, 5 mM sodium citrate (pH 7.0), 0.1 M bmercaptoethanol, and 0.5% sarkosyl]. RNA was extracted from the lysate by CsCl2 gradient centrifugation, and ethanol precipitation. Equal activity (2.56106 c.p.m.) of labelled RNA was used as probes to hybridize with plasmids containing b-actin (5 mg) or RB (10 mg) cDNAs which had been linearized, denatured, and immobilized on nitrocellulose membranes. The hybridization and washing steps afterwards were performed as reported (Greenberg and Bender, 1992).

Acknowledgements The work has been supported by NIH grant (CA 64602). The authors would like to acknowledge Ruthie Lapushin and Tim E Walch for their technical assistance and Tatsuro Furui for his help with preparation of this manuscript.

References Bast RC, Jr, Berek JS, Obrist R, Griths CT, Berkowitz RS, Hacker NF, Parker L, Lagasse LD and Knapp RC. (1983). Cancer Res., 43, 1395 ± 1401. Berchuck A, Rodriguez G, Olt G, Whitaker R, Boente MP, Arrick BA, Clarke-Pearson DL and Bast RC, Jr. (1992). Am. J. Obstet. Gynecol., 166, 676 ± 684. Bremner R, Cohen BL, Sopta M, Hamel PA, Ingles CJ, Gallie B and Phillips RA. (1995). Mol. Cell. Biol., 15, 3256 ± 3265. Buick RN, Pullano R and Trent JM. (1985). Cancer Res., 45, 3668 ± 3676. Caldas C, Hahn SA, daCosta LT, Redston MS, Schutte M, Seymour AB, Weinstein CL, Hruban RH, Yeo CL and Kern SE. (1994). Nature Genet., 8, 27 ± 32. Chen P-L, Scully P, Shew J-Y, Wang JYJ. and Lee WH. (1989). Cell, 58, 1193 ± 1198. Chirgwin JM, MacDonald AERJ and Rutter WJ. (1979). Biochemistry, 18, 5294 ± 5299. Dodson MK, Cliby WA, Xu H-J, DeLacey KA, Hu S-X, Keeney GL, Li J, Podratz KC, Jenkins RB and Benedict WF. (1994). Cancer Res., 54, 610 ± 613. Gill RM, Hamel PA, Jiang Z, Zacksenhaus E, Gallie BL and Phillips RA. (1994). Cell Growth & Dier., 5, 467 ± 474. Gonzalez-Zulueta M, Bender CM, Yang AS, Nguyen TD, Beart RW, Van Tornout JM and Jones PA. (1995). Cancer Res., 55, 4531 ± 4535. Goodrich DW, Wang NP, Qian YW, Lee EYHP and Lee WH. (1991). Cell, 67, 293 ± 302. Graham FL and Prevec L. (1991). Methods in Molecular Biology, Gene Transfer and Expression Protocols. Murray EJ (ed.). Humana Press: Totowa, NJ, pp. 109 ± 128. Greenberg ME, and Bender TP. (1992). Short protocols in Molecular Biology. Ausubel FME (ed.). Greene Publishing Associates and John Wiley & Sons, Inc.: New York. 4.25 ± 4.29. He J, Allen JR, Collins VP, Allalunis-Turner MJ, Godbout R, Day III RS and James CD. (1994). Cancer Res., 54, 5804 ± 5807. He J, Olson JJ and James CD. (1995). Cancer Res., 55, 4833 ± 4836. Hengstschlager M, Hengstschlager-Ottnad E, Pusch O and Wawra E (1996). Oncogene, 12, 1635 ± 1643.

Horowitz JM, Park SH, Bogenmann E, Cheng JC, Yandell DW, Kaye FJ, Minna JD, Dryja TP and Weinberg RA. (1990). Proc. Natl. Acad. Sci. USA, 87, 2775 ± 2779. Jen J, Harper JW, Bigner SH, Bigner DD, Papadaopoulos N, Markowitz S, Willson JKV, Kinzler KW and Vogelstein B. (1994). Cancer Res., 54, 6353 ± 6358. Jiang W, Kahan S, Tomita N, Zhang Y, Lu S and Weinstein B. (1992). Cancer Res., 52, 2980 ± 2983. Jin X, Nguyen D, Zhang WW, Kyritsis AP and Roth JA. (1995). Cancer Res., 55, 3250 ± 3253. Johnson DG. (1995). Oncogene, 11, 1685 ± 1692. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavitian SV, Stockert E, Day R, III, Johnson BE and Skolnik MH. (1994). Science, 264, 430 ± 440. Kamb A. (1995). Cell-cycle regulators and Cancer. Trends Genet., 11, 136 ± 140. Kelley MJ, Nakagawa K, Steinberg SM, Mulshine JL, Kamb A and Johnson BE. (1995). J. Natl. Cancer Inst., 87, 756 ± 761. Laiho M, DeCaprio JA, Ludlow JW, Livingston DM and Massague J. (1990). Cell, 62, 175 ± 185. Lammie G, Fantl V, Smith R, Schuuring E, Brookes S, Michalides R, Dickson C, Arnold A and Peter G. (1991). Oncogene, 6, 439 ± 444. Lengyel P, Choubey D, Li S-J and Datta B. (1995). Semin. Virol., 6, 203 ± 213. Li Y, Nichols MA, Shay JW and Xiong Y. (1994). Cancer Res., 6078 ± 6082. Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss M, Peters G and Bartek J. (1995). Nature, 375, 503 ± 506. Mori T, Miura K, Aoki T, Nishihara T, Mori S and Nakamura Y. (1994). Cancer Res., 54, 3396 ± 3397. Motokura T, Bloom T, Kim G, Juppner H, Ruderman JV, Kronenberg HM and Arnold A. (1991). Nature, 350, 512 ± 515. Nobori T, Miura K, Wu DJ, Lois A, Takbayashi K and Carson DA. (1994). Nature, 368, 753 ± 756. Okamoto A, Demetric DJ, Spillare EA, Hagiwara K, Hussain SP, Bennett WP, Forrester K, Gerwin B, Serrano M, Beach DH and Harris CC. (1994). Proc. Natl. Acad. Sci. USA, 91, 11045 ± 11049. Oswald F, Lovec H, Moroy T and Lipp M. (1994). Oncogene, 9, 2029 ± 2036.

7


8

Otterson GA, Kratzke RA, Coxon A, Kim YW and Kaye FJ. (1994). Oncogene, 9, 3375 ± 3378. Pales E, Ben-Levy R, Tzahar E, Liu N, Wen D and Yarden Y. (1993). EMBO J., 12, 961 ± 971. Parry D, Bates S, Mann DJ and Peters G. (1995). EMBO J., 14, 503 ± 511. Peng HQ, Hogg D, Bailey D, Gallie BL, Bulbul M, Jewett M, Buchanan J and Goss PE. (1993). Cancer Res., 53, 3574 ± 3578. Rodabaugh KJ, Biggs RB, Qureshi JA, Barrett AJ, Welch WR, Bell DA, Berkowitz RS and Mok SC. (1995). Oncogene, 11, 1249 ± 1254. Schmidt EE, Lchimura K, Reifenberger G and Collins VP. (1994). Cancer Res., 54, 6321 ± 6324. Schulze A, Zerfab K, Spitkovsky D, Henglein B and JansenDurr P. (1994). Oncogene, 9, 3475 ± 3482. Serrano M, Hannon GL and Beach D. (1993). Nature, 366, 704 ± 707. Shan B, Chang CY, Jones D and Lee WH. (1994). Mol. Cell. Biol., 14, 299 ± 309. Shapiro GI, Edwards CD, Kobzik L, Godleski J, Richards W, Sugarbaker DJ and Rollins B. (1995a). Cancer Res., 55, 505 ± 509.

Shapiro GI, Park JE, Edwards CD, Mao L, Merlo A, Sidransky D, Ewen ME and Rollins BJ. (1995b). Cancer Res., 55, 6200 ± 6209. Spruck CH, Gonzalez-Zulueta M, Shibata A, Simoneau AR, Lin MF, Gonzales F, Tsai YC and Jones PA. (1994). Nature, 370, 183 ± 184. Ueki K, Ono Y, Henson JW, E®rd JT, von Deimling A and Louis DN. (1996). Cancer Res., 56, 150 ± 153. Weinberg RA. (1995). Cell, 81, 323 ± 330. Xu H-J, Hu S-X and Benedict WF. (1991a). Oncogene, 6, 1139 ± 1146. Xu H-J, Hu S-X, Cagle PT, Moore GE and Benedict, W. F. (1991b). Cancer Res., 51, 2735 ± 2739. Xu L, Sgroi D, Sterner CJ, Beauchamp RL, Pinney DM, Keel S, Ueki K, Rutter JL, Buckler AJ, Louis DN, Gusella JF and Ramesh, V. (1994). Cancer Res., 54, 5262 ± 5264. Zhou X, Tarmin L, Yin J, Jiang HY, Suzuki H, Rhyu MG, Abraham J and Meltzer SJ. (1994). Oncogene, 9, 3737 ± 3741.